Population Mortality Rates Prediction Based On Spatial-temporal Vector Autoregressive Model

The study of population mortality has always been a subject of great social concern,with important implications for the insurance industry,social welfare programs,and other related areas.This thesis employs the concept of "space" from spatial statistics and econometrics to construct a population mortality prediction model based on the Nearest Spatio-Temporal Autoregressive(NSTAR)model,and further analyzes the impact of Covid-19 on population mortality.The first part of this thesis focuses on the construction,empirical analysis,and application of the NSTAR model.Firstly,the structure of the model and the meanings of its components are introduced,and the inclusion of age and cohort effects in the model is illustrated with examples.Additionally,the persistence of residual terms in the model over multiple future periods was analyzed,and the dynamic impact of the interdependence between residual terms on mortality was further demonstrated through simulation experiments.Based on this,the stationarity of the model is proven,and the parameters are estimated using penalized least squares,with hyperparameters selected through cross-validation.Furthermore,the NSTAR dual-population model and multipopulation model are constructed,and specific analyses of the impact paths and parameter estimates of residual terms in the model are conducted.Secondly,empirical analysis is conducted using demographic data from the Human Mortality Database,and the NSTAR model is applied to both single and multi-population cases.The predicted results are compared with the corresponding benchmark models to obtain the optimal forecasting model.Finally,the application value of the NSTAR model is elucidated,and the model is used to calculate future population mortality rates and average life expectancy.The second part of this thesis investigates the impact of Covid-19 on changes in population mortality rates.Firstly,the data used is introduced and described through descriptive statistical analysis.Secondly,changes in life expectancy and life expectancy inequality during the Covid-19 period are studied,as well as the changes in their relationship.Finally,using linear least squares regression,the impact of income on life expectancy is analyzed,and the changes in the influence of income on male and female life expectancy during the Covid-19 period are explored.